Each is a fact about how we work, with the date it was last checked and the place you can check it yourself.
Two published studies are why. University of St. Gallen researchers, arXiv 2604.07585 (April 2026) report a standard error of 0.370 for a single run and recommend at least seven runs per prompt per day. SparkToro with Gumshoe.ai (January 2026) had 600 volunteers run 12 prompts a combined 2,961 times and reports a less than 1 in 100 chance that ChatGPT or Google's AI returns the same list of brands in any two responses.